Gene Therapy Using TRAIL-Secreting Human Umbilical Cord Blood–Derived Mesenchymal Stem Cells against Intracranial Glioma

Introduction

Glioblastomas are the most common primary malignant brain tumors in humans. The prognosis of patients with malignant gliomas is extremely poor because gliomas are refractory to conventional therapies, such as extensive surgical resection, radiation, and chemotherapy ( 1). Therapeutic gene delivery with viral vectors by direct injection into the primary brain tumor or postoperative tumor cavity has failed to reach outgrowing tumor islands because of the migratory abilities of the tumor cells and their infiltration into the normal brain parenchyma ( 2). New effective therapeutic tools are needed that specifically target tumor cells, especially those cells that have escaped the main tumor mass.

Recent studies suggest that stem cells can be used as vehicles for delivering therapeutic genes to treat brain tumors ( 3– 5). Neural stem cells exhibit extensive tropism for experimental gliomas and migrate toward outgrowing microsatellites ( 6– 9). However, the clinical application of neural stem cells is limited by ethical problems associated with their isolation and immunologic incompatibility in allogenic transplantation. Additionally, a new therapeutic strategy has been developed that uses mesenchymal stem cells (MSC) for the targeted delivery and local production of biological agents in tumors ( 10, 11). These reports imply that MSCs are particularly attractive for clinical use because they have tumor targeting properties, can be easily isolated and expanded to the numbers required for use, and can be genetically manipulated with viral vectors.

Human umbilical cord blood (UCB) is an alternative source of adult stem cells. Several studies indicate that UCB-derived MSCs (UCB-MSC) are similar to stem cells from bone marrow with respect to cell characteristics and multilineage differentiation potential ( 12– 15). In a previous study, we isolated multipotent UCB-MSCs that have the capacity to differentiate into several mesodermal tissues (bone, cartilage, tendon, muscle, and adipose), endodermal tissue (hepatocyte), and ectodermal tissue (neurons; refs. 16, 17). Recently, UCB-MSCs were proved to be more advantageous in cell procurement, storage, and transplantation than bone marrow–derived MSCs ( 18). Moreover, the number and differentiation ability of bone marrow–derived MSCs significantly decrease with age ( 19). Cells in UCB, or neonatal blood, are less mature than adult cells and they do not trigger an immense immune reaction in unrelated donor transplantation ( 12, 20). These characteristics make UCB-MSCs potent candidates for the clinical application of allogenic MSC–based therapies.

Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)–based cancer therapies, which involve treatment with recombinant TRAIL (rTRAIL) or an adenovirus bearing the TRAIL gene, against glioma have been shown ( 21, 22); however, clinical trials of these therapies are limited. There are potential issues of toxicity and protein half-life in patients treated with high-dose rTRAIL, and in adenoviral gene therapy, limitations arise from the relatively short survival time of the virus caused by an immune reaction ( 23) and by outgrowing glioma cells infiltrating the brain parenchyma. The artificial TRAIL gene [secretable trimeric TRAIL (stTRAIL)], which encodes a fusion protein composed of three functional elements, a secretion signal, a trimerization domain, and an apoptosis-inducing moiety of the TRAIL gene sequence, has been developed ( 24). Adenoviral vectors delivering the stTRAIL gene (Ad-stTRAIL) showed higher tumor suppressor activity than adenoviral vectors delivering the full-length TRAIL gene in vitro and in vivo ( 25). To extend the release time of stTRAIL and deliver it to infiltrating tumor cells, we evaluated UCB-MSCs secreting stTRAIL as delivery vehicles for brain tumor therapy.

In addition, most of the replication-deficient adenoviral vectors that have been used to transduce MSCs are based on human adenovirus serotype 5 (Ad5). Cell entry of Ad5-based vectors is mediated through a receptor-mediated biphasic process that involves primary attachment to the cellular coxsackie-adenovirus receptor (CAR; ref. 26) and internalization via interaction with integrins present in target cells ( 27). However, transduction of MSCs by conventional Ad5 vectors is inefficient even when very high multiplicities of infection (MOI) are used because MSCs do not express CAR ( 28). Therefore, the development of methods that achieve comparable adenoviral gene delivery with substantially lower viral doses is highly desirable to eventually develop therapies for the treatment of human glioma. One way to increase cell entry of adenovirus vectors may be through cell-permeable peptides [protein transduction domain (PTD)], small polybasic peptides derived from the transduction domains of certain proteins that cross the cell membrane through a receptor-independent mechanism ( 29– 31). These cell-permeable peptides have been used to introduce biologically active cargo molecules, such as DNA, peptides or proteins, and viruses, into cells. Recently, we developed a novel PTD (4HP4), which could significantly enhance adenoviral transduction into MSCs ( 32).

In the present study, we show that human UCB-MSCs display tropism for human glioma and that the treatment of stTRAIL-secreting UCB-MSCs has significant antitumor effects compared with adenoviral TRAIL gene therapy.

Materials and Methods

Culture of human UCB-MSCs and other cell lines. Human UCB harvest and expansion of MSCs isolated from UCB were conducted as previously reported ( 16). The separated MSCs were subcultured at a concentration of 5 × 10⁴/cm² in α-MEM (Invitrogen) and used for experiments during passages 5 to 8. U-87MG, U-251MG, A172, NIH3T3, and 293 cells were obtained from the American Type Culture Collection. U-87MG and 293 cells were maintained in Eagle's MEM (Invitrogen), and U-251MG, A172, and NIH3T3 cells in DMEM (Invitrogen). Normal human astrocytes were obtained from the Applied Cell Biology Research Institute and cultured in DMEM. Enhanced green fluorescent protein (EGFP)–expressing U-87MG cells (U87-EGFP) were derived from stable transfection of U-87MG cells with pEGFP-N1 vector (BD Biosciences) and the stable transfectants were selected with neomycin analogue G418 (Invitrogen). All media were supplemented with 2 mmol/L l-glutamine, 100 units/mL penicillin, 100 μg/mL streptomycin, and 10% fetal bovine serum (FBS) purchased from Invitrogen. Cells were incubated at 37°C in a humidified atmosphere containing 5% CO₂.

Western blotting. UCB-MSCs or 293 cells were lysed in a radioimmunoprecipitation assay (RIPA) buffer (Sigma) containing a protease inhibitor cocktail (Roche). Proteins were separated on a 10% SDS-PAGE and transferred onto a nitrocellulose transfer membrane (Whatman). After blocking the membrane with TBS-0.1% Tween 20 containing 5% skim milk for 1 h at room temperature, the membrane was incubated with rabbit anti-human CAR antibody (Santa Cruz Biotechnology) overnight at 4°C. The membrane was incubated for 1 h at room temperature with horseradish peroxidase–conjugated secondary antibody (Santa Cruz Biotechnology) and the bands were detected using Amersham enhanced chemiluminescence detection reagents (GE Healthcare Life Sciences).

Adenoviral vectors and transduction conditions. The recombinant replication-deficient adenoviral vector encoding the gene for EGFP (Ad-EGFP) was constructed and produced using the Ad-Easy vector system, following the manufacturer's instructions (Quantum Biotechnologies). Adenovirus carrying the stTRAIL gene (Ad-stTRAIL) was engineered as described previously ( 25). Ad-ψ5 was used as a control. Peptide (4HP4) was synthesized by Peptron. ⁶ To transfect UCB-MSCs or U-87MG cells, adenoviruses at a specified MOI were pretreated with 4HP4 in serum-free medium for 30 min at room temperature, and then cells were incubated with the premixed virus-4HP4 complex for 30 min, washed twice with PBS, and changed to the original growth medium. EGFP expression was analyzed by flow cytometry using the FACSCalibur system (Becton Dickinson Co.) or a fluorescence microscope (Axiovert 200, Carl Zeiss).

Animals and brain tumor model. Male athymic nude mice (6–8 wk old; Charles River Laboratories) were used in accordance with institutional guidelines under the approved protocols. For the intracranial xenografts of human glioma, animals were anesthetized with ketamine/xylazine i.p. and stereotactically inoculated with 1 × 10⁵ U-87MG or U87-EGFP cells (in 3 μL PBS) into the right frontal lobe (2 mm lateral and 1 mm anterior to bregma, at 2.5 mm depth from the skull base) via a Hamilton syringe (Hamilton Company) using a microinfusion pump (Harvard Apparatus).

In vitro migration studies. The migratory ability of UCB-MSCs was determined using Transwell plates (Corning Costar) that were 6.5 mm in diameter with 8-μm pore filters. Cells were incubated at three different concentrations of (1 × 10⁵, 5 × 10⁵, and 1 × 10⁶) in serum-free medium for 48 h and the resulting conditioned media were used as chemoattractants. MSCs or MSCs-stTRAIL cells (2 × 10⁴) were suspended in serum-free medium containing 0.1% bovine serum albumin (Sigma) and seeded into the upper well; 600 μL of conditioned medium were placed in the lower well of the Transwell plate. Following incubation for 5 h at 37°C, cells that had not migrated from the upper side of the filters were scraped off with a cotton swab, and filters were stained with the Three-Step Stain Set (Diff-Quik; Sysmex). The number of cells that had migrated to the lower side of the filter was counted under a light microscope with five high-power fields (×400). Experiments were done in triplicate.

In vivo migration studies. Seven days after tumor cell inoculation into the right frontal lobe, EGFP-expressing MSCs (MSCs-EGFP; 2 × 10⁵) were implanted into the opposite hemisphere (2 mm lateral and 1 mm anterior to bregma, at a depth of 2.5 mm from the skull base), or PKH26 (Sigma)–labeled MSCs-stTRAIL were implanted into the tumor mass. Migration toward the tumor was assessed at 4, 7, or 10 d after MSCs inoculation by direct visualization with a fluorescence microscope or a confocal microscope (LSM 510 Meta, Carl Zeiss).

Assessment of cell viability and terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling staining. UCB-MSCs or U-87MG cells (5 × 10⁴) were seeded in 24-well plates, and increasing amounts of Ad-stTRAIL or recombinant human TRAIL (rhTRAIL; R&D Systems) were added to confirm TRAIL tumor-specific cytotoxicity. At 2 or 3 d after treatment, cells were analyzed for viability by the 3-(4,5-dimethyl-thiazol-2yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) assay (Promega). For coculture experiments, MSCs-stTRAIL or U87-stTRAIL cells were plated in the Transwell inserts containing 0.4-μm pores (Corning Costar) with increasing cell concentrations (0.2 × 10⁴, 0.5 × 10⁴, 1 × 10⁴, 2 × 10⁴, and 4 × 10⁴ per well) and then U-87MG cells (2 × 10⁴) were grown in the lower well of the Transwell plates. For inhibition studies, MSCs-stTRAIL (2 × 10⁴) were seeded in the upper well and U-87MG cells (2 × 10⁴) were seeded in the lower well of the Transwell plates, and then neutralizing antihuman TRAIL antibody (R&D Systems) was added to the lower wells as indicated in the figure. After 5 d, the viability of U-87MG cells in the lower well was analyzed by MTS assay. All experiments were conducted in triplicate. To detect apoptotic activity, MSCs-stTRAIL and U-87MG cells (2 × 10⁴ each) were premixed and cultured in four-well chamber slides (Nalge Nunc International) for 48 h and then stained using a terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) assay kit (Roche) developed with Cy3-conjugated streptavidin (Jackson ImmunoResearch Laboratories).

Flow cytometry for TRAIL death receptors. Cells were analyzed for the surface expression of TRAIL death receptors with phycoerythrin-conjugated antihuman DR4, DR5, DcR1, and DcR2 (R&D Systems). Briefly, cells (2.5 × 10⁵) were stained with each antibody on ice for 30 min. After washing with PBS, the expressions of these death receptors were analyzed by flow cytometry using the FACSVantage SE (Becton Dickinson Co.).

ELISA for expressed stTRAIL. stTRAIL protein secreted into the culture supernatants or brain tissues was analyzed by ELISA, as described previously ( 25). To investigate the persistence of transgene expression in vitro, UCB-MSCs or U-87MG cells were seeded at a high density (4 × 10⁴ per well of a 24-well plate) and transduced with Ad-stTRAIL. The virus-containing medium was removed and additionally incubated in low-serum medium (α-MEM or Eagle's MEM containing 2% FBS). Culture supernatants were harvested and changed with fresh medium every 3 d, and secreted TRAIL was assessed at various time intervals. For the analysis of stTRAIL expressed in tumor-bearing mice in vivo, tumor tissues were harvested and lysed in RIPA buffer at 1, 4, 7, 10, and 14 d after treatment of MSCs-stTRAIL.

Treatment of mouse experimental glioma. To evaluate the therapeutic effects of MSCs-stTRAIL in vivo, a single intratumoral (i.t.) injection was conducted at 7 d after tumor inoculation. Tumor-bearing mice were injected with MSCs-EGFP or MSCs-stTRAIL (2 × 10⁵ cells in 5 μL PBS), PBS (5 μL), and Ad-stTRAIL (1 × 10¹⁰ viral particles; this is safe in mice and also a therapeutically effective viral dose) for the survival experiment. To assess inhibition of tumor growth, tumor-bearing mice were injected with unmodified MSCs or MSCs-stTRAIL (2 × 10⁵ cells in 5 μL of PBS) and treated with PBS (5 μL).

Evaluation of tumor size by histologic analysis. Tumor size was determined as described previously ( 33). Briefly, brains with therapeutic treatment at a specific time point after tumor inoculation were serially sectioned (20 μm, obtained every 200 μm into the tumor) and then stained with H&E. The section with the maximum tumor area was identified and the number of pixels in the delineated area was calculated via a computer using the NIH Image software. ⁷ The pixel count was then normalized to the original dimensions (square millimeter) of the scanned sections.

Immunohistochemistry and in vivo apoptosis assay. Mouse brains were perfused with PBS followed by 4% paraformaldehyde under deep anesthesia at a specific time point after MSCs-stTRAIL inoculation. The excised brains were postfixed overnight and then equilibrated in PBS containing 30% sucrose for 2 d. Fixed brains were embedded, snap frozen in liquid nitrogen, and stored at −70°C until use. Tissues were cryosectioned (14 μm) and then stained with primary antibodies for anti-TRAIL (R&D Systems) or antihuman nuclei (Chemicon) using the Vector M.O.M. Immunodetection Kit (Vector Laboratories). The primary antibodies were detected either Cy3- or fluorescein-conjugated streptavidin (Jackson ImmunoResearch Laboratories). To detect apoptotic activity, TUNEL staining was done as described above. In some sections, nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; Sigma) for counterstaining.

Statistical analysis. All data are expressed as mean ± SE. Statistical differences between different test conditions were determined by Student's t test. P < 0.05 was considered significant. Statistical analysis of survival was done using log-rank test.

Results

Effect of PTD on transduction efficiency of UCB-MSCs with adenovirus. Human primary MSCs are relatively resistant to wild-type adenoviral infection because of their low expression level of the adenoviral receptor CAR. Thus, we first confirmed whether human UCB-MSCs used in this study expressed CAR protein. The expression of CAR from UCB-MSCs was not detected by Western blot analysis ( Fig. 1A ). Therefore, we conducted PTD-mediated adenoviral transduction as a CAR-independent infection method, and the gene delivery efficiency of Ad-stTRAIL into UCB-MSCs was determined ( Fig. 1B, a). We found that transduction of UCB-MSCs could be greatly enhanced by incorporation of 4HP4 (0.003–0.03 μmol/L) into the Ad-stTRAIL infection medium. However, along with increased MOI and 4HP4 concentration, cell death was also augmented, possibly reflecting adenovector cytopathogenic effects and/or cell damage due to excessive transgene expression ( Fig. 1B, b). Taken together, a 4HP4 concentration of 0.01 μmol/L and relatively low doses of adenoviruses achieved high-level production of proteins without affecting cell viability and rendered optimal transduction efficiency of UCB-MSCs. Additionally, a significant dose-dependent increase in the number of GFP-positive cells was detected, and enhanced transduction efficiency was evident on visual inspection when the Ad-EGFP infection was carried out in the presence of 0.01 μmol/L 4HP4 (data not shown). In addition, the persistence of transgene expression in MSCs-stTRAIL or U87-stTRAIL cells (U-87MG cells transduced with Ad-stTRAIL under the same conditions as UCB-MSCs; however, the transduction efficiency of U-87MG was not affected by incorporation of 4HP4) was studied under conditions of restrained cell division. ELISA analysis at different time points after transduction revealed that the concentration of secreted TRAIL remained fairly constant for 23 days and then began to decline in MSCs-stTRAIL; however, a decrease occurred in U87-stTRAIL cells after day 8 ( Fig. 1C). Based on these results, we used cells engineered under these transduction conditions (20 MOI of viruses plus 0.01 μmol/L of 4HP4 in infection medium) in all subsequent experiments.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Effect of 4HP4 on adenoviral transduction into UCB-MSCs. A, CAR expression was determined by Western blot analysis. Cell lysate from the 293 cell line served as a positive control. B, the effect of 4HP4 on transduction efficiency of UCB-MSCs infected with Ad-stTRAIL was evaluated. Cells were transduced at various MOIs without or with increasing concentrations of 4HP4. At day 3 after infection, the concentration of secreted TRAIL in culture supernatant was analyzed by ELISA (a) and the viability of transduced cells was analyzed via MTS assay (b). Columns and points, mean; bars, SE. C, the persistence of transgene expression in UCB-MSCs or U-87MG cells transduced with Ad-stTRAIL (MOI 20) containing 4HP4 (0.01 μmol/L) was analyzed under conditions of restrained cell division. The concentration of secreted TRAIL was assessed by ELISA at each time point. Columns, mean; bars, SE.

Migratory capacity of UCB-MSCs toward gliomas in vitro and in vivo. It has been reported that a factor released from glioma cells may be a potential chemoattractant involved in the tropism of MSCs. To test this in UCB-MSCs, in vitro migration assays using Transwell plates were conducted. Conditioned medium from normal human astrocytes was used as a control to better mimic the normal brain milieu. Only a few cells migrated toward serum-free medium and conditioned medium from fibroblasts or normal human astrocytes, whereas the migration of MSCs-stTRAIL was significantly (P < 0.001) stimulated by conditioned medium from human glioma cell lines (U-87MG, U-251MG, or A172) compared with conditioned medium from NIH3T3 cells or astrocytes ( Fig. 2A ). Moreover, the migratory activity of MSCs-stTRAIL was represented in a dose-dependent manner ( Fig. 2B). We also confirmed that unmodified MSCs migrated to the conditioned medium from human glioma cells in a similar pattern to MSCs-stTRAIL (data not shown). These results indicated that human glioma cells were capable of stimulating the migration of MSCs and that the migratory ability of UCB-MSCs was not affected by adenoviral transduction.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Migratory ability of transduced UCB-MSCs in vitro. A, the migratory ability of UCB-MSCs in response to conditioned medium (CM) from different cell lines was determined using a Transwell plate (8-μm pores). Representative photomicrographs of stained filters show migrated MSCs-stTRAIL. Magnification, ×100. B, migration of MSCs-stTRAIL in response to three different concentrations of conditioned medium from three human glioma cell lines (U-87MG, U-251MG, and A172), NIH3T3 cells, or normal human astrocytes was determined. Cell migration was compared and evaluated after staining by taking photographs and counting cells that had migrated under a light microscope. Serum-free medium (SFM) was used as a negative control (horizontal line). Columns, mean; bars, SE. *, P < 0.001, Student's t test.

Next, we investigated whether implanted MSCs could migrate toward intracranial gliomas in vivo. MSCs-EGFP inoculated into the contralateral hemisphere to the tumor migrated away from the initial injection site toward the tumor mass along the corpus callosum at 7 days after MSCs inoculation ( Fig. 3A, a–d ), whereas MSCs-EGFP remained within the injection site in the normal brain ( Fig. 3A, e–g). These cells were mostly retained at the corpus callosum and the border between the tumor and normal parenchyma; they also infiltrated into the tumor bed. We also confirmed that MSCs-stTRAIL inoculated into the contralateral hemisphere migrated toward the tumor mass (data not shown). More importantly, PKH26-labeled MSCs-stTRAIL injected directly into the tumor bed distributed extensively throughout the tumor mass at 4 or 10 days after MSCs inoculation ( Fig. 3B). Although the MSCs-stTRAIL remained within the injection site, they largely migrated at the border of the tumor where it interfaced with normal tissue ( Fig. 3B, b and c). Interestingly, these cells were also seen in tumor satellites along the corpus callosum at a distance from the main tumor mass ( Fig. 3B, d–g). We showed that the migration of MSCs-stTRAIL occurs within 4 days, with appreciable levels of TRAIL secretion being present after i.t. administration (see Fig. 6A), and the migrating human MSCs remain at the border of the tumor where it interfaced with normal tissue at day 10. We also confirmed that the mock-transduced or nontransduced cells migrated toward intracranial gliomas in a similar pattern to transduced MSCs (data not shown). Thus, this result showed that UCB-MSCs have an excellent migratory capacity and glioma tropism in vivo.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Migration of UCB-MSCs toward gliomas in vivo. MSCs-EGFP or PKH26-labeled MSCs-stTRAIL cells were implanted into the opposite hemisphere to the U-87MG tumor (A) or i.t. into the U87-EGFP tumor mass (B). A, MSCs-EGFP (green) cells were implanted into the opposite hemisphere to the tumor in glioma-bearing mice (a–d) or normal mice as a control (e–g). At day 7 after implantation, MSCs-EGFP were seen through the corpus callosum (b, box 1 from a) and in the tumor mass (c, box 2 from a). d, H&E staining shows the interface of the tumor and normal brain in a contiguous section of c. MSCs-EGFP were only seen at the injection site (f, box 1 from e) and were not found within the corpus callosum (g, box 2 from e) or the opposite hemisphere in normal mice. T, tumor mass; dotted line, tumor edge; arrowheads, MSCs-EGFP cells located in the tumor mass; insets, higher magnification (×400) of the area indicated by arrow. Magnification, ×200. B, PKH26-labeled MSCs-stTRAIL cells (red) were implanted into the tumor mass (green) in an established glioma (a, H&E staining). At day 4 after implantation, PKH26-labeled MSCs-stTRAIL cells can be seen extensively interspersed among the green tumor cells (b, box 1 from a) or distributed throughout the tumor mass (c, box 2 from a), and also seen in a tumor satellite (f, box 4 from a) through the corpus callosum (d and e, box 3 from a). A higher magnification (×400) of boxed area in d shows the migrating MSCs (e). The cells stained with human nuclei antibody (hNA; green) of boxed area in f revealed the injected human MSCs (g, ×800). At day 10 after implantation, MSCs-stTRAIL were mostly to be found at the border between tumor and normal parenchyma (h and i). H&E staining shows the interface of the tumor and normal brain (h, ×100). T, primary tumor; t, tumor satellite; arrow, injection site of PKH26-labeled MSCs-stTRAIL. Magnification, ×200. Nuclei were stained with DAPI (blue) for counterstaining.

TRAIL induces apoptosis in U-87MG cells but not in UCB-MSCs. To quantitatively assess the cytotoxic effect of TRAIL on U-87MG cells and UCB-MSCs, we cultured these cells in media containing various concentrations of rhTRAIL and infected both cell types with an increasing MOI of Ad-stTRAIL. There was a significant dose-dependent decrease in the viability of U-87MG cells that were supplemented with rhTRAIL or infected with Ad-stTRAIL (P < 0.01). In contrast, UCB-MSCs did not show any decrease in viability with exposure to rhTRAIL protein or TRAIL gene transfer, even at high concentrations or MOIs ( Fig. 4A ). To investigate the cause of resistance to TRAIL in UCB-MSCs, the cell surface expression of TRAIL receptors possessing a death domain (DR4 and DR5) or not (DcR1 and DcR2) was evaluated by flow cytometry ( Fig. 4B). U-87MG cells, sensitive to TRAIL, expressed high levels of DR5, the major receptor involved in TRAIL-induced apoptosis. DR5 was also detected in UCB-MSCs but at lower levels compared with the expression in U-87MG cells. UCB-MSCs expressed significant levels of DcR1 decoy receptor that have high affinity to TRAIL, suggesting that the overexpression of DcR1 blocks DR5 functions in TRAIL-induced apoptosis by competing with DR5 for TRAIL. We then cultured premixed MSCs-stTRAIL and U-87MG cells to determine whether TRAIL secretion from transduced MSCs could induce apoptosis in U-87MG cells. This resulted in significant apoptosis in U-87MG cells premixed with MSCs-stTRAIL; however, U-87MG cells premixed with Ad-ψ5-infected MSCs (MSCs-ψ5) did not undergo apoptosis ( Fig. 4C, a). In addition, neutralization of secreted TRAIL from transduced MSCs fully abolished apoptosis induction of U-87MG cells in coculture system using Transwell plates, suggesting that the killing is mediated by secreted TRAIL ( Fig. 4C, b). These findings confirmed that MSCs-stTRAIL secreted biologically relevant quantities of TRAIL protein, which affected only transformed cells without damaging normal cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Tumor-specific cytotoxicity of TRAIL and therapeutic efficacy of MSC-based gene delivery in vitro. A, U-87MG cells and UCB-MSCs were infected with different doses of Ad-stTRAIL or treated with varying concentrations of rhTRAIL. Two days after rhTRAIL treatment or 3 d after Ad-stTRAIL infection, cell viability was analyzed using an MTS assay. *, P < 0.01, Student's t test. B, the cell surface expression of TRAIL death receptors in U-87MG cells and UCB-MSCs was analyzed by flow cytometry (green lines). Mouse isotype IgG1 antibody served as a control (black lines). Results are representative of three independent determinations for each receptor and cell line. Percent of positive cells was determined. C, U-87MG cells premixed with MSCs-stTRAIL were cultured for 48 h and then analyzed using the TUNEL assay. A coculture of MSCs-ψ5 and U-87MG cells was used as a negative control. TUNEL-positive nuclei (red) were stained and counterstaining was conducted with DAPI (blue; a). Magnification, ×200. Secreted TRAIL–mediated apoptosis was confirmed by blocking with neutralizing antihuman TRAIL antibody in coculture of MSCs-stTRAIL and U-87MG cells using Transwell plates containing semiporous membranes (0.4-μm pores). After 5 d, the viability of U-87MG cells in the lower well was assayed by MTS (b). Control, 2 × 10⁴ cells of MSCs-ψ5 in the upper wells. Columns, mean; bars, SE. D, effect of MSCs-stTRAIL compared with U87-stTRAIL on survival of U-87MG cells was determined using Transwell plates (0.4-μm pores). After 5 d, the viability of U-87MG cells in the lower well was assayed by MTS (a) and the concentration of secreted TRAIL was determined using an ELISA assay (b). Control, 4 × 10⁴ cells of MSCs-ψ5 or U87-ψ5 in the upper wells. Columns and points, mean; bars, SE. *, P < 0.05; **, P < 0.001, Student's t test.

Therapeutic potential of TRAIL-secreting UCB-MSCs as delivery vehicles in vitro. To determine the therapeutic benefit of MSCs-stTRAIL compared with U87-stTRAIL cells, which could mimic the effects of injecting the adenovirus encoding stTRAIL gene to the tumor mass directly, U-87MG cells were cocultured with MSCs-stTRAIL or U87-stTRAIL. To determine whether this growth inhibition was specifically due to the release of soluble TRAIL, we used Transwell plates containing semiporous membranes that separated the cells. Whereas treatment with MSCs-stTRAIL resulted in a dose-dependent inhibition of U-87MG cell growth, treatment with U87-stTRAIL was limited in its effects on tumor cell death ( Fig. 4D, a). To verify that this effect was due to the release of TRAIL, the concentration of TRAIL in the medium was determined ( Fig. 4D, b). A dose-dependent increase in the concentration of TRAIL directly correlated with tumor cell death in the MSCs-stTRAIL treatment group; however, an increase in TRAIL concentration was not observed in the U87-stTRAIL treatment group and there was no correlation with increasing cell numbers. This was because transduced U-87MG cells were killed by the TRAIL secreted from them, which correlates with the results shown in Fig. 1C; thus, they could not continuously secret the therapeutic gene.

Effects of MSCs-stTRAIL on tumor growth and survival of glioma-bearing mice in vivo. To assess whether inoculation of MSCs-stTRAIL showed antitumor effects in vivo, maximal tumor surface areas were determined by histologic analysis from glioma-bearing mice. The average maximal tumor areas in MSCs-stTRAIL–treated animals were decreased compared with PBS- or MSCs-treated animals. This decrease in tumor size associated with MSCs-stTRAIL treatment was significant at day 14 (P = 0.007, MSCs-stTRAIL versus MSCs; P = 0.0016, MSCs-stTRAIL versus PBS) and at day 21 (P = 0.022, MSCs-stTRAIL versus MSCs; P = 0.039, MSCs-stTRAIL versus PBS; Fig. 5A ). There was no detectable difference in tumor size between animals treated with MSCs and PBS. These results show that stTRAIL-secreting MSCs, when i.t. implanted in established glioma, reduce the rate of tumor growth.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Effects of MSCs-stTRAIL on tumor growth and survival of glioma-bearing mice in vivo. A, MSCs-stTRAIL or MSCs (2 × 10⁵ cells) and PBS were given i.t. to glioma-bearing mice at day 7 after U-87MG (10⁵ cells) inoculation. a, tumor size was determined by histologic analysis at 14 or 21 d after tumor inoculation (n = 5 per treatment group). Columns, mean; bars, SE. *, P < 0.01; **, P < 0.05, Student's t test. b, representative photographs of H&E staining from each group. Magnification, ×1. B, survival curve of intracranial glioma-bearing mice. At day 7 after U-87MG (10⁵ cells) inoculation, tumors were injected with a single dose (2 × 10⁵ cells) of MSCs-stTRAIL (n = 10) or MSCs-EGFP (n = 10) and inoculated with Ad-stTRAIL (1 × 10¹⁰ viral particles; n = 10). The PBS-treated group was used as a control (n = 10). Analysis of survival was conducted by a log-rank test based on the Kaplan-Meier method. Representative result of three independent experiments.

Next, the survival of MSCs-stTRAIL–treated mice was significantly prolonged compared with controls treated with PBS or MSCs-EGFP ( Fig. 5B). There were no detectable differences in survival among animals treated with MSCs-EGFP, unmodified MSCs (data not shown), or PBS. Statistical analysis revealed that the effects of MSCs-stTRAIL and Ad-stTRAIL were both significantly different from that of MSCs-EGFP (P = 0.0001, MSCs-stTRAIL versus MSCs-EGFP; P = 0.001, Ad-stTRAIL versus MSCs-EGFP). However, MSCs-stTRAIL–treated mice showed prolonged survival compared with Ad-stTRAIL–treated mice (P = 0.0241), which was evaluated for the comparison between MSC-based therapy and viral gene therapy at a single i.t. injection. These data indicated that the treatment of stTRAIL-secreting UCB-MSCs has a strong antitumor effect compared with adenoviral TRAIL gene therapy.

In vivo expression of TRAIL and induction of apoptosis in established intracranial brain tumors by inoculated MSCs-stTRAIL. We quantitatively measured the levels of stTRAIL protein and its longevity in vivo to determine whether the antitumor activity of MSCs-stTRAIL correlated with the levels of stTRAIL expressed in tumor tissues. ELISA analysis on different days after treatment revealed that the TRAIL protein expressed in tumor tissues was detectable on day 1, peaked on day 4, began to decrease after day 7, and persisted for 2 weeks ( Fig. 6A ). In the control group (MSCs-EGFP), TRAIL protein was not detected throughout the experimental period. Additionally, immunohistochemical analysis of tissues prepared from mice at day 7 showed strong staining for TRAIL within inoculated tumors and on the border of tumors, indicating the presence of TRAIL-secreting human UCB-MSCs ( Fig. 6B). Furthermore, to evaluate the apoptosis-inducing ability of TRAIL secreted from UCB-MSCs in vivo, TUNEL staining was done ( Fig. 6C). MSCs-stTRAIL–treated tumors were almost completely apoptotic, and apoptosis was confined to the tumor mass and not normal brain parenchyma, indicating that MSCs-stTRAIL inoculated i.t. was inducing tumor cell death. Apoptotic activity detected in MSCs-EGFP– or PBS-treated tumors was negligible. Importantly, apoptotic cells were detected in the main tumor mass and were seen in proximity to invading tumor islands, indicating that MSCs-stTRAIL migrated through the outgrowing tumor from the primary tumor site into adjacent normal tissue.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

TRAIL expression and apoptosis in MSCs-stTRAIL–treated gliomas in vivo. At day 7 after U-87MG (10⁵ cells) inoculation, MSCs-stTRAIL or MSCs-EGFP (2 × 10⁵ cells) were treated i.t. A, for the quantification of stTRAIL levels and longevity of stTRAIL expression in tumor tissues, brain tissues were homogenized at 1, 4, 7, 10, and 14 d (n = 3 per group) after treatment and then assessed by ELISA. Control, MSCs-EGFP–treated brain tissues. Columns, mean; bars, SE. B, at day 7 after MSCs-stTRAIL inoculation, treated brains were stained for TRAIL. a, H&E staining shows the interface of the tumor and normal brain (×100). b, sections from MSCs-stTRAIL–treated brains showed positive staining for TRAIL (red) in the tumor mass and at the tumor border line, indicating the presence of MSCs-stTRAIL (×400). c to f, a higher magnification image of the TRAIL-stained cells (arrows) illustrated in b shows that the TRAIL-positive cell nuclei were stained with human nuclei antibody (hNA, green; ×800). Nuclei were stained with DAPI (blue) for counterstaining. Arrowheads, human nuclei antibody–stained cell nuclei. C, MSCs-stTRAIL–treated brains stained with TUNEL (red) show the specific staining of apoptosis in the tumor and the lack of staining in adjacent normal tissue (a); however, a section from MSCs-EGFP–treated brains shows negligible staining (b; ×100). c, apoptotic cells were detected in the main tumor mass (×200); inset, a high-power view (×800) of TUNEL-stained nuclei from the area indicated with arrow. d, additionally, apoptotic cells were seen in close proximity to invading tumor islands (×200). Nuclei were stained with DAPI (blue) for counterstaining. T, tumor mass; dotted line, tumor edge; arrowheads, tumor island.

Discussion

In this study, we efficiently engineered human MSCs derived from UCB to secret stTRAIL via adenoviral transduction mediated by cell-permeable peptides and provided evidence that these cells can migrate toward human gliomas. Additionally, considerable reduction in tumor growth and prolongation of survival occur in glioma-bearing mice treated with TRAIL-secreting UCB-MSCs.

We used a xenogenic model of human glioblastoma grown in nude mice for evaluating the migratory ability and the therapeutic efficacy of human MSCs. Recently, it has been reported that the human xenograft glioblastoma models in immunodeficient mice exhibit histopathologic features compatible with tumor invasion into the normal/nonneoplastic brain parenchyma, although the tumor borders in the murine models are not as diffuse as those of the spontaneous glioblastoma in human ( 34). Therefore, murine models of glioblastoma seem to recapitulate several of the human glioblastoma histopathologic features, and considering their reproducibility and availability, they constitute a valuable in vivo system for preclinical studies.

The migration of UCB-MSCs in vitro was stimulated by conditioned medium from cultured glioma cells in a dose-dependent manner. This indicates that soluble factors released from glioma cells could induce the migration of MSCs. It is known that these factors released from cancer cells promote the recruitment of endothelial cells and endogenous stromal cells from the bone marrow toward the tumor mass, and such responses closely resemble tissue remodeling after injury or inflammation ( 35, 36). Similar mechanisms would be mediated in the migration of UCB-MSCs in glioma; however, these tumor-specific migratory properties require further elucidation in relation to their potential use in therapeutic applications. In our in vivo experiments, we found that when injected contralaterally or i.t. to the tumor mass, UCB-MSCs show a tracking ability through outgrowing glioma cells. Although this is consistent with reports describing the tumor-specific migratory abilities of MSCs derived from bone marrow ( 11), UCB-MSCs have many advantages, such as the immaturity of newborn cells compared with adult cells, large ex vivo expansion capacity, low risk of viral infection, lack of donor attrition, and less pronounced immune response. Thus, it is important that the strategies using the tracking ability of UCB-MSCs to achieve widespread distribution of therapeutic agents throughout outgrowing gliomas could improve brain tumor therapy.

In the present study, we engineered UCB-MSCs expressing TRAIL, which selectively induce apoptosis in a wide variety of transformed cells without damaging normal cells and tissues ( 37), to investigate the therapeutic potential of MSCs delivering proapoptotic proteins. UCB-MSCs could be used as therapeutic vehicles to deliver TRAIL because they are resistant to TRAIL-mediated apoptosis, which would extend the release time of our stTRAIL gene and could lead to delivery of the therapeutic gene to outgrowing tumor cells because of the cell tropism for gliomas. The coculture experiments in vitro showed that treatment of MSCs-stTRAIL is more effective than U87-stTRAIL treatment, which mimics adenoviral gene therapy that is injected directly into the tumor mass. Furthermore, the survival experiment in vivo revealed that direct injection of Ad-stTRAIL did not result in enhanced survival compared with treatment of MSCs-stTRAIL at a single dose of i.t. administration. These results indicate that MSC-based TRAIL gene therapy is more advantageous than viral gene therapy, which involves direct injection of an adenovirus-encoding TRAIL gene into tumors. However, further work needs to be done in relation to the safety and brain toxicity of soluble TRAIL at high local concentrations before this MSCs-stTRAIL becomes a candidate for a human clinical glioma therapy trial.

In agreement with a study showing that the efficiency of exogenous gene transfer into MSCs using adenoviral vectors is relatively low ( 28), we confirmed that UCB-MSCs lack CAR expression and show poor transduction of exogenous genes. In flow cytometry analysis, when UCB-MSCs were infected with Ad-EGFP, about 20% to 30% of cells were transduced, even at a high MOI (300–500). At higher MOIs (≥1,000), ∼80% of cells became transduced, suggesting that adenovirus entry occurs via low-affinity interactions with nonspecific attachment molecules; however, we found that cell viability was severely hampered by viral cytopathogenic effects (data not shown). To circumvent this problem, several groups have reported that fiber-modified adenoviral vectors can result in expanded tropism compared with wild-type adenoviral vectors ( 38, 39). In the present study, we engineered TRAIL-secreting UCB-MSCs by PTD-mediated adenoviral transduction using a novel PTD (4HP4), which improves adenoviral gene expression at reduced titers of the virus in various cells in vitro and the efficacy of therapeutic gene transduction in vivo ( 32). Our data show that UCB-MSCs can be efficiently transduced by adding 4HP4 to the infection medium (∼90% represented in flow cytometry analysis) and migration capacity was not affected by PTD-mediated adenoviral transduction. Therefore, addition of 4HP4 during adenoviral infection of MSCs can enhance the transduction efficiency, facilitating the use of adenoviral vectors in MSC-mediated gene therapies.

Recently, it has been reported that MSCs could support tumor growth ( 40) and promote cancer metastasis ( 41). Although these finding suggest that MSCs could favor tumor growth under physiologic conditions, the antitumor effects of TRAIL secreted from engineered MSCs can lead to reversal of tumor growth in our therapy. Moreover, the supportive effect of MSCs alone on efficacy experiments in the present study was not observed, which implies that the proportion of inoculated UCB-MSCs, which reached only about 5% in established tumor masses after a single injection, was insufficient to support tumor growth.

Although prolonged survival of glioma-bearing mice by treatment with MSCs-stTRAIL was shown, we did not achieve complete tumor regression by single i.t. administration. We found that the expression from MSCs-stTRAIL inoculated into glioma-bearing mice persisted for 14 days but began to decline after 7 days. This is one of the possible explanations for incomplete tumor regression in relation to the elimination of engineered UCB-MSCs and the progressive growth of glioma cells. These results suggest that further investigations on the repeated administration of MSCs-stTRAIL with controlled doses and appropriate time intervals and the expression level of stTRAIL in tumor tissues will be needed for the clinical application. Systemic administration has the advantage that repeated injections are clinically feasible. Although systemic administration by tail vein injection has been reported for mouse neural stem cells ( 42), we found that the human MSCs were filtered by the lung and few cells were detected in the intracranial tumor of nude mice (data not shown), which may have been due to species incompatibilities ( 11). In addition, a wide surgical resection of tissue in malignant gliomas is highly limited because this might result in an unwanted loss of brain function. Therefore, a certain amount of tumor mass might remain in the primary site after surgical resection. Thus, it will be useful to evaluate the combination of surgery, which is accompanied by other therapies such as chemotherapy and radiation therapy, and repeated MSCs-stTRAIL therapy in human patients.

In conclusion, we have shown for the first time the migratory capacity of UCB-MSCs toward gliomas and show an efficient adenoviral transduction system for MSCs. The engineered UCB-MSCs secreting stTRAIL are effective in regressing tumor growth in the intracranial xenograft mouse model. Thus, the use of UCB-MSCs as delivery vehicles of therapeutic genes will be of great interest for the clinical application of stem cell–based cancer therapy.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.